Evolution of expert systems for real-time process management: A case study on motor control

Abstract

Expert systems are rapidly moving into real-time process control applications, usually in conjunction with distributed digital control systems. These applications include intelligent alarm management, operator advisories for start-up/shut-down/monitoring functions, and process and equipment diagnostics. In most current systems, the expert systems are implemented using stand-along computer systems, which may be either dedicated LISP processors or conventional computers. This paper discusses the evolution of on-line system implementations from the stand-alone computer approach to imbedded processors within the distributed control architecture. The paper summarizes the benefits that result from this evolution, including a reduction in system hardware and engineering costs, ergonomic benefits in operator interfacing, and a significant improvement in system performance. In addition, those benefits are demonstrated in a case study in the area of motor control. In this case study, an existing set of Boolean logic for motor control is also implemented in expert system production rules. It was found that the production rule approach offers two major advantages: First, the production rules are more understandable; second, production rules are more easily tested because of the features of the expert system shell.